1 |
148 |
jeremybenn |
/*******************************************************************************
|
2 |
|
|
*
|
3 |
|
|
* Copyright (c) 1993 Intel Corporation
|
4 |
|
|
*
|
5 |
|
|
* Intel hereby grants you permission to copy, modify, and distribute this
|
6 |
|
|
* software and its documentation. Intel grants this permission provided
|
7 |
|
|
* that the above copyright notice appears in all copies and that both the
|
8 |
|
|
* copyright notice and this permission notice appear in supporting
|
9 |
|
|
* documentation. In addition, Intel grants this permission provided that
|
10 |
|
|
* you prominently mark as "not part of the original" any modifications
|
11 |
|
|
* made to this software or documentation, and that the name of Intel
|
12 |
|
|
* Corporation not be used in advertising or publicity pertaining to
|
13 |
|
|
* distribution of the software or the documentation without specific,
|
14 |
|
|
* written prior permission.
|
15 |
|
|
*
|
16 |
|
|
* Intel Corporation provides this AS IS, WITHOUT ANY WARRANTY, EXPRESS OR
|
17 |
|
|
* IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTY OF MERCHANTABILITY
|
18 |
|
|
* OR FITNESS FOR A PARTICULAR PURPOSE. Intel makes no guarantee or
|
19 |
|
|
* representations regarding the use of, or the results of the use of,
|
20 |
|
|
* the software and documentation in terms of correctness, accuracy,
|
21 |
|
|
* reliability, currentness, or otherwise; and you rely on the software,
|
22 |
|
|
* documentation and results solely at your own risk.
|
23 |
|
|
*
|
24 |
|
|
* IN NO EVENT SHALL INTEL BE LIABLE FOR ANY LOSS OF USE, LOSS OF BUSINESS,
|
25 |
|
|
* LOSS OF PROFITS, INDIRECT, INCIDENTAL, SPECIAL OR CONSEQUENTIAL DAMAGES
|
26 |
|
|
* OF ANY KIND. IN NO EVENT SHALL INTEL'S TOTAL LIABILITY EXCEED THE SUM
|
27 |
|
|
* PAID TO INTEL FOR THE PRODUCT LICENSED HEREUNDER.
|
28 |
|
|
*
|
29 |
|
|
******************************************************************************/
|
30 |
|
|
|
31 |
|
|
.file "strncat.s"
|
32 |
|
|
#ifdef __PIC
|
33 |
|
|
.pic
|
34 |
|
|
#endif
|
35 |
|
|
#ifdef __PID
|
36 |
|
|
.pid
|
37 |
|
|
#endif
|
38 |
|
|
/*
|
39 |
|
|
* (c) copyright 1988,1993 Intel Corp., all rights reserved
|
40 |
|
|
*/
|
41 |
|
|
/*
|
42 |
|
|
procedure strncat (optimized assembler version for the 80960K Series)
|
43 |
|
|
|
44 |
|
|
dest_addr = strncat (dest_addr, src_addr, max_bytes)
|
45 |
|
|
|
46 |
|
|
append the null terminated string pointed to by src_addr to the null
|
47 |
|
|
terminated string pointed to by dest_addr. Return the original
|
48 |
|
|
dest_addr. If the source string is longer than max_bytes, then
|
49 |
|
|
append only max_bytes bytes, and tack on a null byte on the end.
|
50 |
|
|
|
51 |
|
|
This routine will fail if the source and destination string
|
52 |
|
|
overlap (in particular, if the end of the source is overlapped
|
53 |
|
|
by the beginning of the destination). The behavior is undefined.
|
54 |
|
|
This is acceptable according to the draft C standard.
|
55 |
|
|
|
56 |
|
|
Undefined behavior will also occur if the end of the source string
|
57 |
|
|
(i.e. the terminating null byte) is in the last two words of the
|
58 |
|
|
program's allocated memory space. This is so because strncat fetches
|
59 |
|
|
ahead. Disallowing the fetch ahead would impose a severe performance
|
60 |
|
|
penalty.
|
61 |
|
|
|
62 |
|
|
Strategy:
|
63 |
|
|
|
64 |
|
|
First, skip to the null byte in the destination string. Then
|
65 |
|
|
fetch the source string by words and store them by words to the
|
66 |
|
|
destination string, until there are fewer than three bytes left
|
67 |
|
|
to copy. Then, using the last word of the source (the one that
|
68 |
|
|
contains the remaining 0, 1, 2, or 3 bytes to be copied), store
|
69 |
|
|
a byte at a time until Ldone.
|
70 |
|
|
|
71 |
|
|
If, before exhausting the max_byte count, the null byte is encountered
|
72 |
|
|
in the source string, then just copy up thru the null byte.
|
73 |
|
|
|
74 |
|
|
Tactics:
|
75 |
|
|
|
76 |
|
|
1) Do NOT try to fetch and store the words in a word aligned manner
|
77 |
|
|
because, in my judgement, the performance degradation experienced due
|
78 |
|
|
to non-aligned accesses does NOT outweigh the time and complexity added
|
79 |
|
|
by the preamble and convoluted body that would be necessary to assure
|
80 |
|
|
alignment.
|
81 |
|
|
*/
|
82 |
|
|
|
83 |
|
|
.globl _strncat
|
84 |
|
|
.globl __strncat
|
85 |
|
|
.leafproc _strncat,__strncat
|
86 |
|
|
.align 2
|
87 |
|
|
_strncat:
|
88 |
|
|
#ifndef __PIC
|
89 |
|
|
lda Lrett,g14
|
90 |
|
|
#else
|
91 |
|
|
lda Lrett-(.+8)(ip),g14
|
92 |
|
|
#endif
|
93 |
|
|
__strncat:
|
94 |
|
|
mov g14,g6
|
95 |
|
|
cmpibge 0, g2, Lno_operation # Lexit early if max_bytes <= 0
|
96 |
|
|
mov g0, g5
|
97 |
|
|
Lskip_word_loop:
|
98 |
|
|
ld (g5), g7 # fetch word of dest string
|
99 |
|
|
addo 4, g5, g5 # post-increment dest ptr
|
100 |
|
|
scanbyte 0, g7 # does it contain null byte?
|
101 |
|
|
bno Lskip_word_loop # if not, loop
|
102 |
|
|
subo 5, g5, g5 # adjust dest ptr
|
103 |
|
|
lda 0xff, g3 # byte extraction mask = 0xff;
|
104 |
|
|
Lskip_byte_loop:
|
105 |
|
|
and g7, g3, g14 # extract byte of last word of dest string
|
106 |
|
|
cmpo 0, g14 # is it null?
|
107 |
|
|
addo 1, g5, g5 # adjust dest ptr
|
108 |
|
|
shro 8, g7, g7 # position next byte for extraction
|
109 |
|
|
bne Lskip_byte_loop # loop if null not found yet
|
110 |
|
|
|
111 |
|
|
ld (g1), g7 # fetch first word of source string
|
112 |
|
|
Lwloop: # word copying loop
|
113 |
|
|
cmpo 4, g2 # max_bytes < 4 ?
|
114 |
|
|
addo 4, g1, g1 # post-increment source ptr
|
115 |
|
|
bge Lcloop.a # branch if less than 4 bytes to move
|
116 |
|
|
scanbyte 0, g7 # is null byte reached yet?
|
117 |
|
|
mov g7, g4 # keep a copy of the source word
|
118 |
|
|
be Lcloop # branch if null byte reached
|
119 |
|
|
ld (g1), g7 # pre-fetch next word of source
|
120 |
|
|
subo 4, g2, g2 # reduce max_byte counter
|
121 |
|
|
st g4, (g5) # store current word
|
122 |
|
|
addo 4, g5, g5 # post-increment destination ptr
|
123 |
|
|
b Lwloop
|
124 |
|
|
|
125 |
|
|
Lcloop.b:
|
126 |
|
|
addo 1, g5, g5 # post-increment destination ptr
|
127 |
|
|
shro 8, g7, g7 # position next byte for extraction
|
128 |
|
|
Lcloop: # character copying loop (max_byte > 3)
|
129 |
|
|
and g3, g7, g4 # extract character
|
130 |
|
|
cmpo 0, g4 # is it null?
|
131 |
|
|
stob g4, (g5) # store it
|
132 |
|
|
bne Lcloop.b # loop if null not encountered yet
|
133 |
|
|
|
134 |
|
|
bx (g6) # g0 = dest string address; g14 = 0
|
135 |
|
|
Lrett:
|
136 |
|
|
ret
|
137 |
|
|
|
138 |
|
|
Lcloop.c:
|
139 |
|
|
addo 1, g5, g5 # post-increment destination ptr
|
140 |
|
|
shro 8, g7, g7 # position next byte for extraction
|
141 |
|
|
Lcloop.a: # character copying loop (max_byte <= 3)
|
142 |
|
|
cmpdeco 0,g2,g2 # max_byte == 0?
|
143 |
|
|
and g3, g7, g4 # extract character
|
144 |
|
|
be Ldone # store null and Lexit if max_byte exhausted
|
145 |
|
|
cmpo 0, g4 # is it null?
|
146 |
|
|
stob g4, (g5) # store it
|
147 |
|
|
bne Lcloop.c # loop if null not encountered yet
|
148 |
|
|
|
149 |
|
|
Ldone: stob g14, (g5) # store trailing null
|
150 |
|
|
bx (g6) # g0 = dest string address; g14 = 0
|
151 |
|
|
|
152 |
|
|
Lno_operation: mov 0, g14 # conform to register conventions
|
153 |
|
|
bx (g6)
|
154 |
|
|
|
155 |
|
|
/* end of strncat */
|